A question about Mosfet clippers

[mike_a]

Hi all,

I'm new to the site, but I've been lurking around for some time now.  Really love the forum!

That said, I've been looking at numerous Mosfet clipping circuits and I'm somewhat confused.
I almost all of them, including Shaka 3 etc., the Mosfets are connected in diode configuration (gate & drain shorted) and a regular diode is connected in series.
The thing that is confusing is that the Mosfet diode and the normal diode are facing opposite directions!
You can see it here:
http://www.diystompboxes.com/wiki/index.php?title=Featuredpedals

This way you don't get the Mosfet in diode configuration.  You get the Substrate-source inherent diode which is like a normal diode.

I would appreciate it if someone could explaion this to me.

Thanks,
Mike.

[R.G.]

It takes a little - but only a little - semiconductor knowledge to understand this.

A MOSFET, like most other semiconductors today, is made like a stack of pancakes. They start with a base (substrate) of lightly-doped (that is, higher resistance; more doping = lower resistance) silicon. This bottom layer is attached to the source of MOSFETs

On top of that they put a layer of opposite-polarity-doped silicon for the bottom part of the active device. This layer is more heaviliy doped, because it's a current carrier for the active device and you want resistive losses lower. This layer becomes the channel, with drain at one end and source at the other end.

For a MOSFET, the next layer is a layer of insulating glass. This is the isolation between the channel, which is what is below it, and the gate which is above. it. The top layer is a layer of metal on the top of the glass to control current flow.

In normal use, the channel-substrate diode is reverse biased. That is, for an N-channel, the source is the lowest voltage and the drain is the highest voltage. The reverse voltage varies from 0 at the source (it's connected by a wire!) to whatever the drain voltage is at the drain. That can be 100's of volts when it's non conducting down to a fraction of a volt when it's fully enhanced.

When the source is made more positive than the rest of the channel, the substrate-channel diode is forward biased, and when it hits the conduction threshold, current starts flowing through it and out the drain. That's the reverse-diode on a MOSFET.

If you connect a MOSFET gate-drain, the nominal "diode connection" for a MOSFET, then the MOSFET is actualy doing its normal job of conducting when the gate is forward biased. It just has a very low drain voltage. It's actually amplifiying, but it is constrained to only do this very near the gate-threshold turn on voltage. The "diode forward voltage" for this connection is equal to the Vthreshold plus a little. The "plus a little" is all of the amplifying region it can do.

So the MOSFET "diode connection" points in the opposite direction from the substrate diode. If we hook this up as a clipper, then it clips like the substrate diode one direction and the MOSFET diode connection the other way. These are wildly different diodes. Presumably there was a reason we wanted to hear the MOSFET diode to go to all this trouble to get, so we would like to isolate only the MOSFET conduction diode, not the substrate diode, which is very much like other silicon diodes.

The way that's done is by putting another diode in series with the diode-connected MOSFET. This other diode is only there as a switch. It is pointed in the way opposite to the substrate diode. That way, when the voltage across the combined external diode and MOSFET is in the direction to make the substrate diode conduct, the external diode blocks it. When the voltage across the combined external diode and MOSFET is in the direction to make the MOSFET amplify-conduct, the external diode lets that current through.

So the external diode selects which of the MOSFET amplifying-mode-diode versus the substrated diode conducts. This lets us get rid of the (presumably ordinary-sounding) substrate diode conduction and use only the MOSFET amplifying-mode-diode for clipping.

It's worth noting that although the substrate diode is an asymmetrically doped silicon diode, it will have a sound all its own, and that may be useful too. There are very few diode-diodes on the market that are as lightly doped, so you may WANT to use just the substrate diode. Some people do. In that case, just hook up two MOSFETs in reverse-parallel, hook gate to drain to protect the gate on each MOSFET and bag the external diodes. All you hear is the substrate diodes because they conduct at a voltage lower than the MOSFET amplifying diode connection on the opposite MOSFET, so all you'll hear is the two substrates.

[mike_a]

First of all, Thank you for the well written and knowlegable reply!

I agree with everything you said.
However, In the schematic contained in the link I attached (and others I've seen), the cathode of the normal diode is pointing in the direction of the source of the Mosfet.
From what you explain (unless I missunderstand), this is exactly the opposite direction to connect it if you want to prevent the substarte body diode.
This way no current will flow when the Mosfet diode is forward biased!

Thanks again,
Mike.

[WGTP]

R.G and John Greene helped some of us thru this a while back and if you look thru my stuff you will see I have put it to us.  Most of the schemes, as you said, use the plain diode and not the Mosfet.  Most Mosfets have a pretty high clipping threshold (BS170/2N7000 around 2v.) and may require special design considerations to get good results.  I like them.  There are several threads that address this if you search.   

http://www.aronnelson.com/gallery/v/WGTP/

[mike_a]

Ok, I got it.

Was starting to think I got everything backwards 
So, the schem I brought in the link actually uses the substarte diode intentionally.

Anyone heard a big difference between this setup and the normal diode setup?

Also, I don't think that a 2v clipping threshold is that bad.  LED's are somewhere in that region, right?

[GuitarLord5000]

Here's a way you can get the same result without using extra diodes:
http://www.muzique.com/lab/zenmos.htm (Thanks Jack!) Look at the second diagram.

And yes, there is a marked difference in this setup vs. regular silicon or germanium diodes.

Due to the much higher voltage drop with this setup, the clipping is much softer than either germanium or silicon.  Also, the 'knee' is supposed to be softer.

[WGTP]

Here is one I have reposted several times.   It covers the subject pretty well. 

[mike_a]

Thanks guys!

You were all really helpful.

[WGTP]

http://www.diystompboxes.com/smfforum/index.php?topic=38581.0 

This is what I meant to post.